2007/7/24, Ivo Emanuel Gon?alves <justivo@gmail.com>:> > On 7/24/07, Greg M. <contrabassboy@yahoo.com> wrote: > > Ivo, Harry is asking about CPU usage of the DEcoder, > > not the ENcoder. > > Sorry, my bad. > > I believe that FLAC's decoding is somewhat faster than most other > lossless formats, as FLAC is a much less complex format.Maybe somebody knows some comparison tables with some measured values like the ones available for encoding times? (so now I'm looking for the decoding CPU power loads) thx _______________________________________________> Flac mailing list > Flac@xiph.org > http://lists.xiph.org/mailman/listinfo/flac >-------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.xiph.org/pipermail/flac/attachments/20070725/e7dd88d0/attachment.htm
Harry, Keep in mind that the processor load will be different for every processor model. PowerPC G4, G5, and then all the implementations of x86. Processor load does not depend upon clock speed - all that clock speed determines is how fast the operation can be done, and particularly whether it can be done in real time. Back to your question: The CPU load will be determined by the efficiency of the instruction set for the processor running the program, as well as how well the compiler maps the C source to those instructions. Any comparison table you might find could be irrelevant if your processor model was not tested, and if the compiler options are changed when you build FLAC, then that might change the efficiency as well. I point this out because FLAC is not always going to be "better" or "worse" than other formats. We're talking about several moving targets here, all of which influence each other. Brian Willoughby Sound Consulting On Jul 25, 2007, at 11:21, Harry Sack wrote: 2007/7/24, Ivo Emanuel Gon?alves <justivo@gmail.com>: On 7/24/07, Greg M. <contrabassboy@yahoo.com> wrote: > Ivo, Harry is asking about CPU usage of the DEcoder, > not the ENcoder. Sorry, my bad. I believe that FLAC's decoding is somewhat faster than most other lossless formats, as FLAC is a much less complex format. Maybe somebody knows some comparison tables with some measured values like the ones available for encoding times? (so now I'm looking for the decoding CPU power loads) thx -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.xiph.org/pipermail/flac/attachments/20070725/83dcf5ad/attachment.html
Harry, Another thing to consider is the balance between CPU efficiency and disk speed. On some of my systems, decoding a FLAC file to AIFF (or WAV) uses 100% of the CPU. That's because the drive is faster than the CPU, so the CPU is constantly working. Moving to a 4-processor system, I can run 4 FLAC decodes at the same time. At first, that would not use 400% (100% of all 4 CPUs) because the disk was not fast enough to read 4 files and write 4 files at once. But as soon as I upgraded to a faster drive on a faster bus, I am back to 400% CPU usage when decoding. I use FLAC to back up original multi-track recordings. Then I burn the FLAC files to CD-R or DVD-R. Whenever a client needs the originals for a mixing session, I have to pull all the FLAC files from optical media and decode them, since nearly all mixing software uses AIFF (even the ones that allow FLAC will convert the files from FLAC to something else on disk). In order to respond to the client as fast as possible, I want to decode the FLAC files as fast as possible, so I have experimented with faster machines and faster drives. In other words, I'm not sure that you can make any solid conclusions from a comparison table, even if you could find one, because it would always be possible to recompile flac, or upgrade the CPU, or upgrade the disk, or add additional disks to spread the load. There is no single answer to your question. Brian Willoughby Sound Consulting On Jul 25, 2007, at 11:21, Harry Sack wrote: Maybe somebody knows some comparison tables with some measured values like the ones available for encoding times? (so now I'm looking for the decoding CPU power loads) thx
Perhaps a more important question is how much electricity is required to decode FLAC, and how much heat is generated - for embedded CPUs of course. FLAC might not become popular for portable players if it shortens their battery life significantly. While there is the problem that .flacs take up more Flash drive space than lossily compressed tracks, portable players with larger capacities are more widely available now, and cheaper than they used to be, so I expect the use of flac on portable players to become more widespread. Best, Mike Crawford aka Rippit the Ogg Frog rippit@oggfrog.com http://www.oggfrog.com/